BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Ajay Mishra

Ajay Mishra appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
astro-ph.SRastro-ph.GAComputation and LanguageMachine Learningeess.ASSound

Catalog footprint

What is connected

5works
6topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

5 published item(s)

preprint2020arXiv

A Multitask Learning Approach for Diacritic Restoration

In many languages like Arabic, diacritics are used to specify pronunciations as well as meanings. Such diacritics are often omitted in written text, increasing the number of possible pronunciations and meanings for a word. This results in a more ambiguous text making computational processing on such text more difficult. Diacritic restoration is the task of restoring missing diacritics in the written text. Most state-of-the-art diacritic restoration models are built on character level information which helps generalize the model to unseen data, but presumably lose useful information at the word level. Thus, to compensate for this loss, we investigate the use of multi-task learning to jointly optimize diacritic restoration with related NLP problems namely word segmentation, part-of-speech tagging, and syntactic diacritization. We use Arabic as a case study since it has sufficient data resources for tasks that we consider in our joint modeling. Our joint models significantly outperform the baselines and are comparable to the state-of-the-art models that are more complex relying on morphological analyzers and/or a lot more data (e.g. dialectal data).

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

TRANS-BLSTM: Transformer with Bidirectional LSTM for Language Understanding

Bidirectional Encoder Representations from Transformers (BERT) has recently achieved state-of-the-art performance on a broad range of NLP tasks including sentence classification, machine translation, and question answering. The BERT model architecture is derived primarily from the transformer. Prior to the transformer era, bidirectional Long Short-Term Memory (BLSTM) has been the dominant modeling architecture for neural machine translation and question answering. In this paper, we investigate how these two modeling techniques can be combined to create a more powerful model architecture. We propose a new architecture denoted as Transformer with BLSTM (TRANS-BLSTM) which has a BLSTM layer integrated to each transformer block, leading to a joint modeling framework for transformer and BLSTM. We show that TRANS-BLSTM models consistently lead to improvements in accuracy compared to BERT baselines in GLUE and SQuAD 1.1 experiments. Our TRANS-BLSTM model obtains an F1 score of 94.01% on the SQuAD 1.1 development dataset, which is comparable to the state-of-the-art result.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

On the Relation between the Mysterious 21 Micrometer Emission Feature of Post-Asymptotic Giant Branch Stars and Their Mass Loss Rates

Over two decades ago, a prominent, mysterious emission band peaking at ~20.1 micrometer was serendipitously detected in four preplanetary nebulae (PPNe; also known as "protoplanetary nebulae"). So far, this spectral feature, designated as the "21 micrometer" feature, has been seen in 18 carbon-rich PPNe. The nature of the carriers of this feature remains unknown although many candidate materials have been proposed. The 21 micrometer sources also exhibit an equally mysterious, unidentified emission feature peaking at 30 micrometer. While the 21 micrometer feature is exclusively seen in PPNe, a short-lived evolutionary stage between the end of the asymptotic giant branch (AGB) and planetary nebula (PN) phases, the 30 micrometer feature is commonly observed in all stages of stellar evolution from the AGB through PPN to PNe phases. We derive the stellar mass loss rates (M_{loss}) of these 21 micrometer sources from their dust infrared (IR) emission, using the "2-DUST" radiative transfer code for axisymmetric dusty systems which allows one to distinguish the mass loss rates of the AGB phase (\dot{M_{AGB}}) from that of the superwind (\dot_{M_{SW}}) phase. We examine the correlation between \dot{M_{AGB}} or \dot_{M_{SW}} and the fluxes emitted from the 21 and 30 micrometer features. We find that both features tend to correlate with \dot{M_{AGB}}, suggesting that their carriers are probably formed in the AGB phase. The nondetection of the 21 micrometer feature in AGB stars suggests that, unlike the 30 micrometer feature, the excitation of the carriers of the 21 micrometer feature may require ultraviolet photons which are available in PPNe but not in AGB stars.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

A Tale of Three Mysterious Spectral Features in Carbon-Rich Evolved Stars: The 21 Micrometer, 30 Micrometer, and "Unidentified Infrared" Emission Features

The mysterious "21 micrometer" emission feature seen almost exclusively in the short-lived protoplanetary nebula (PPN) phase of stellar evolution remains unidentified since its discovery two decades ago. This feature is always accompanied by the equally mysterious, unidentified "30 micrometer" feature and the so-called "unidentified infrared" (UIR) features at 3.3, 6.2, 7.7, 8.6, and 11.3 micrometer which are generally attributed to polycyclic aromatic hydrocarbon (PAH) molecules. The 30 micrometer feature is commonly observed in all stages of stellar evolution from the asymptotic giant branch (AGB) through PPN to the planetary nebula phase. We explore the interrelations among the mysterious 21 micrometer, 30 micrometer, and UIR features in the Galactic and Magellanic Cloud of the 21 micrometer sources. We derive the fluxes emitted in the observed UIR, 21 micrometer, and 30 micrometer features from published ISO or Spitzer/IRS spectra. We find that none of these spectral features correlate with each other. This argues against a common carrier (e.g., thiourea) for both the 21 micrometer feature and the 30 micrometer feature (otherwise these two features should correlate). This also does not support large PAH clusters as a possible carrier for the 21 micrometer feature.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Probing the Role of Carbon in the Interstellar Ultraviolet Extinction

We probe the role of carbon in the ultraviolet (UV) extinction by examining the relations between the amount of carbon required to be locked up in dust [C/H]_dust with the 2175 Angstrom extinction bump and the far-UV extinction rise, based on an analysis of the extinction curves along 16 Galactic sightlines for which the gas-phase carbon abundance is known and the 2175 Angstrom extinction bump exhibits variable strengths and widths. We derive [C/H]_dust from the Kramers-Kronig relation which relates the wavelength-integrated extinction to the total dust volume. This approach is less model-dependent since it does not require the knowledge of the detailed optical properties and size distribution of the dust. We also derive [C/H]_dust from fitting the observed UV/optical/near-infrared extinction with a mixture of amorphous silicate and graphite. We find that the carbon depletion [C/H]_dust tends to correlate with the strength of the 2175 Angstrom bump, while the abundance of silicon depleted in dust shows no correlation with the 2175 Angstrom bump. This supports graphite or polycyclic aromatic hydrocarbon (PAH) molecules as the possible carrier of the 2175 Angstrom bump. We also see that [C/H]_dust shows a trend of correlating with 1/R_V, where R_V is the total-to-selective extinction ratio, suggesting that the far-UV extinction is more likely produced by small carbon dust than by small silicate dust.

0 citations0 reviewsNo code linkedOpen access